Sony Q3 FY2009 Earnings Announcement

[NathansFortune]

I think FFXIII will do better on PS3 WW, but equal in NA.

Which pre-order charts are you referring to, COMGNET are for Japan only. Anyway, the talk of FFXIII doing better on 360 was way back when PS3 was more expensive. Now they are equally priced and PS3 offers more in the way of value.

I mean you're not going to buy an Arcade console and play an RPG on it having to restart every time you turn off and changing discs when for a bit more money you can get PS3 version which is superior in graphics and has only one disc and has an unlimited number of save slots.

So basically I think FFXIII will have a larger effect for PS3.

Also Sony mention they expect 15% cost reductions over the coming financial year which puts them from ~ $315 ($295 sale price to retailers, "6 cents in the dollar lost") currently to ~ $267 per unit. This could allow them to bundle Arc for the current price and have a core package without Arch for $249. I'm not sure if they will go for the $249 PS3, but it is looking pretty clear that MS will bundle Natal into the Elite package and avoid a direct price cut this year (due to the expense of Natal) and the cost reductions give Sony an effective way to keep profitability and market Arc effectively against Natal.

My own feeling is that Sony will introduce the real PS3 slimline in 2011, and this will be the real cost cutting unit which they will sell for $199 and until then they will be playing a hold on for dear life strategy rather than the aggressive one we saw this year.

 

[Carl B]

My own feeling is that Sony will introduce the real PS3 slimline in 2011, and this will be the real cost cutting unit which they will sell for $199 and until then they will be playing a hold on for dear life strategy rather than the aggressive one we saw this year.

I don't get people thinking that new systems equal cost-cutting efforts. It's the other way around; cost-cutting efforts equal new systems. Whether the PS3 gets released in a different form factor or not, any and all avenues of cost reduction will be incorporated in any event. The form-factor reduction is simply a means of bundling together some of the incremental marginal improvements allowed through the major improvements that take place regularly.

If we assume that Cell and (perhaps) RSX are being fabbed at 32nm come 2011... and that chip inventories are low enough on 45nm to warrant their immediate inclusion, we just have to think whether the total system footprint would be small enough to warrant a new casing design. The savings from those die reductions, PSU capacity reductions, heat sink lightening, ongoing BD maturation, etc etc would make their way to the system regardless.

 

[deepbrown]

I don't get people thinking that new systems equal cost-cutting efforts. It's the other way around; cost-cutting efforts equal new systems. Whether the PS3 gets released in a different form factor or not, any and all avenues of cost reduction will be incorporated in any event. The form-factor reduction is simply a means of bundling together some of the incremental marginal improvements allowed through the major improvements that take place regularly.

If we assume that Cell and (perhaps) RSX are being fabbed at 32nm come 2011... and that chip inventories are low enough on 45nm to warrant their immediate inclusion, we just have to think whether the total system footprint would be small enough to warrant a new casing design. The savings from those die reductions, PSU capacity reductions, heat sink lightening, ongoing BD maturation, etc etc would make their way to the system regardless.

Correct. There were lots of system changes before the Slim. And I expect there will be lots of changes to the Slim itself.

 

[Carl B]

On FFXIII I think the PS3 version will do better as well. That's not due to any analysis of multiple discs or anything else - I just think that on the Great Console Venn Diagram, there is more overlap amongst FF 'fans' and Playstation 'fans.' That's my monster analysis on that one.

PS - I know its weird but the thread should be about more than just PS3 sales vs 360 sales and where FFXIII is going to sell.

Listened to the conference call yesterday - (I hate Sony's translated calls the most of any company) - and I think things look generally promising for them on a couple of fronts. They seem to have the TV business on an upward trajectory; costs have been cut and they feel they have a good product mix for 2010. Their target is 20M+ sets, but they were hesitant to say whether they felt they would be profitable if they fall short of that number - obviously, a lot just comes down to price competition over the next year.

Vaio's did great as we know, but going forward volatility in the memory markets and such may decrease their margins. Their B2B production/services efforts have hit a slow spell due to reduced corporate spending, but BD is growing. This is a major focus of theirs going forward though, and I feel they are banking on theater upgrades, broadcasting deals, and encoding/production work (in 3D and otherwise) to grow that business.

Music and Movies did well enough, two words: Michael Jackson.

Financial Services obviously doing much better than a year ago - profitable again for one, with much more activity. Sony-Ericsson JV not doing well at all though; they're going to have to have their own Motorola style product shift for sure.

 

[NathansFortune]

I don't get people thinking that new systems equal cost-cutting efforts. It's the other way around; cost-cutting efforts equal new systems. Whether the PS3 gets released in a different form factor or not, any and all avenues of cost reduction will be incorporated in any event. The form-factor reduction is simply a means of bundling together some of the incremental marginal improvements allowed through the major improvements that take place regularly.

If we assume that Cell and (perhaps) RSX are being fabbed at 32nm come 2011... and that chip inventories are low enough on 45nm to warrant their immediate inclusion, we just have to think whether the total system footprint would be small enough to warrant a new casing design. The savings from those die reductions, PSU capacity reductions, heat sink lightening, ongoing BD maturation, etc etc would make their way to the system regardless.

Well Sony themselves are saying just 15% for FYE 31/3/11 which is very low considering the cost has come down an average of 25-30% a year so far. My reasoning behind it is much like yours, I think by 2011 much of the silicon will be fabbed at 32nm or lower and it will make sense both from a hardware and marketing point of view to release a redesigned console. A lot of people still see the PS3 as too large.

 

[JPT]

A report by The WSJ claims that Sony is losing just $18 per PS3 rather than the $37 loss they were making per unit in December last year. Apparently, the company is optimistic that they will make a profit on the hardware in 2011.

Is from this link http://www.maxconsole.net/?mode=news&newsid=38486 but when you click the WSJ link on that page, there is no mention of the USD 18. But WSJ is a pay site right, because the article looks like its cut short.

 

[Crossbar]

My reasoning behind it is much like yours, I think by 2011 much of the silicon will be fabbed at 32nm or lower and it will make sense both from a hardware and marketing point of view to release a redesigned console.

So far Sony has had a two year life cycle on each process for the Cell and the RSX, with the exception of the first year where Cell was replaced with a shrinked version within one year. The RSX has not yet hit the 45 nm process in retail, so it´s highly unlikely that we will see the RSX on the 32 nm process in 2011.

Still if we get a 32 nm Cell and a 45 nm RSX together with new power efficient 1 Gbit XDR dram circuits that may be enough to allow a smaller case, but the cost of the case itself is probably very small, so they may keep the current one in order to not to confuse the market. The move to the PS3 Slim case was very strategic. I am not sure they want to make another move within just two years from that unless they have some new strategic goals they want to achieve, like repositioning the console in the market once more.

 

[AlNom]

If we assume that Cell and (perhaps) RSX are being fabbed at 32nm come 2011...

Someone refresh me on current die sizes please 32 or 28nm might be way too small for RSX itself but maybe you meant a Cell-RSX fusion?

Even assuming crap node scaling of say... 30% instead of the ideal 50%, 1/3rd of the 90nm die size is still rather small. There's also the interface to the XDR to consider too. hm....

 

[Crossbar]

Even assuming crap node scaling of say... 30% instead of the ideal 50%, 1/3rd of the 90nm die size is still rather small. There's also the interface to the XDR to consider too. hm....

By using the Elpidas 1 Gbit XDR DRAM at twice the current speed they could reduce the number of data pins with 50% while keeping the current memory bandwidth.

I think the Cell would be OK with the current XDR setup at 32 nm, but going smaller may need either reducing the pins in some clever way or merging with RSX.

 

[Carl B]

Is from this link http://www.maxconsole.net/?mode=news&newsid=38486 but when you click the WSJ link on that page, there is no mention of the USD 18. But WSJ is a pay site right, because the article looks like its cut short.

The Maxconsole thing is wrong in the literal sense; Sony said 6 cents on the dollar, not 18 dollars per console. One might think them equivalent, but they are not, and even the 6 cents on the dollar figure in my mind is not a fixed number - though it is a snapshot in time at least.

 

[Carl B]

Someone refresh me on current die sizes please 32 or 28nm might be way too small for RSX itself but maybe you meant a Cell-RSX fusion?

Actually I wasn't considering packaging at all... packaging will find a way - that's what I always say!

Cell+RSX combo I see as not happening either, unless there is a serious redesign effort to work out the memory controller discrepancy. Though I guess they could simply do two dies, one package, two buses? Seems cumbersome and not sure how that final silicon would look.

Even assuming crap node scaling of say... 30% instead of the ideal 50%, 1/3rd of the 90nm die size is still rather small. There's also the interface to the XDR to consider too. hm....

The die savings have been a bit mediocre thus far, but 32nm should be on a metal gates process; maybe that scope of transition will have prompted a further re-architecting to remap the I/O? The level of effort given towards it might dovetail with whatever Sony might be doing in terms of the PS4 CPU... but then again, it might be completely free floating as well.

 

[Silent_Buddha]

By using the Elpidas 1 Gbit XDR DRAM at twice the current speed they could reduce the number of data pins with 50% while keeping the current memory bandwidth.

That would require a redesign of the chip. Which costs money. Whether the money spent would be less than the potential money saved is unknown. But either way a redesign of the chip is rather unlikely.

Regards,
SB

 

[Crossbar]

That would require a redesign of the chip. Which costs money. Whether the money spent would be less than the potential money saved is unknown. But either way a redesign of the chip is rather unlikely.

The point of moving to a smaller process is to save money mainly because
A: manufacturing will get cheaper.
B: the chip will use less power to some degree.

If the cost saves don´t match the cost of moving to a new process when looking at the planned volumes, they will not move to the new process. The minor redesign mentioned above would be a small cost taken into account in that decision making.
The required redesign of the (pin) layout of the Cell when moving from 45 to 32 nm will probably be a much larger redesign.

 

[NathansFortune]

Actually I wasn't considering packaging at all... packaging will find a way - that's what I always say!

Cell+RSX combo I see as not happening either, unless there is a serious redesign effort to work out the memory controller discrepancy. Though I guess they could simply do two dies, one package, two buses? Seems cumbersome and not sure how that final silicon would look.



The die savings have been a bit mediocre thus far, but 32nm should be on a metal gates process; maybe that scope of transition will have prompted a further re-architecting to remap the I/O? The level of effort given towards it might dovetail with whatever Sony might be doing in terms of the PS4 CPU... but then again, it might be completely free floating as well.

I don't see RSX+Cell combo chip like EE+GS happening either, they use different memory controllers...

I was under the impression that the reason Cell doesn't scale down well is because of the Flex I/O, which is something they can't move off-die. Unless they can address a new 1Gbit XDR chip with fewer lanes at the same bandwidth how will this not continue to be a problem? Maybe a higher clock can help solve the problem, but this could have wider ranging issues and cause lower yields (which is no good if you are trying to achieve a cost saving).

I think maybe RSX+Cell on the same package would be possible, but it would have a difficult arrangement as they address different types of memory, plus there could be heat issues even at 32nm as the Cell doesn't seem to scale down all that well.

Still though, if anyone can reduce costs it's Sony. Their engineers have proved themselves very capable of producing reliable hardware at low cost with the Slim and even the Phat to a certain degree, they can't be blamed for the high cost of components early on, that's a design issue...

 

[NavNucST3]

...
I mean you're not going to buy an Arcade console and play an RPG on it having to restart every time you turn off and changing discs when for a bit more money you can get PS3 version which is superior in graphics and has only one disc and has an unlimited number of save slots.

So basically I think FFXIII will have a larger effect for PS3.

....

I do not disagree that the PS3 version will outsell the 360 version but your comparison is flawed. There has never been a version of the "Arcade" that has not had at least a memory card. Now, the Arcade has on-board memory for saves and in the US is still quite a bit cheaper than the Elite and PS3.

 

[Carl B]

I was under the impression that the reason Cell doesn't scale down well is because of the Flex I/O, which is something they can't move off-die. Unless they can address a new 1Gbit XDR chip with fewer lanes at the same bandwidth how will this not continue to be a problem?

It has been a problem thus far primarily because the shrinks have been 'dumb' shrinks to a large extent. With the metal gate transition at 32nm should they choose (or really rather, need) to spend effort on the porting of Cell to the new process, they might go the extra step to re-optimize the layout of the chip such that the I/O is no longer creating an artificially high lower bound to the die size. After all at a certain point the benefits do warrant the costs.

But, they might just go with a quicky port to 32nm as well. I was just exploring both sides of the issue.

The talk of single-package was born out of Alstrong's pin-count observations... going to denser XDR could assist there as well, but again it's just going to be about costs to re-architect vs per unit savings across x number of consoles. Who knows where those figures line up these days.

 

[Crossbar]

I don't see RSX+Cell combo chip like EE+GS happening either, they use different memory controllers...

Memory controllers can switched if you can emulate the same or slightly better bandwidth and latency.

Benefits of exchanging the GDDR3 with XDR would be less heat and less memory circuits by using Elpidas 1 Gbit circuits which has several data width configurations and possibly by using circuits with even higher densities going forward. GDDR3 will not offer the same possibilities.

At introduction XDR DRAM was possibly significantly more expensive than GDDR3 DRAM, according to the last i-Supply cost break down of the PS3, XDR DRAM is pretty much on par with other DRAM types. The cost reduction from reducing the number of circuits may favour XDR DRAM.

 

[NathansFortune]

I do not disagree that the PS3 version will outsell the 360 version but your comparison is flawed. There has never been a version of the "Arcade" that has not had at least a memory card. Now, the Arcade has on-board memory for saves and in the US is still quite a bit cheaper than the Elite and PS3.

I thought after NXE the NAND included in the Arcade was reserved.

I'm not really up with Arcade consoles since I have a 60GB Pro, but I know someone who bought a HDless 360 and was pissed they couldn't save games and had to buy a HD.

 

[Silent_Buddha]

I thought after NXE the NAND included in the Arcade was reserved.

I'm not really up with Arcade consoles since I have a 60GB Pro, but I know someone who bought a HDless 360 and was pissed they couldn't save games and had to buy a HD.

I believe pre-NXE arcades are the only ones that had problems due to having to have NXE installed on the HDD. Thus MS had the special program for Arcade owners at the time to pick up a specially priced 20 gig HD for 20 USD.

Regards,
SB

 

[Silent_Buddha]

I thought after NXE the NAND included in the Arcade was reserved.

I'm not really up with Arcade consoles since I have a 60GB Pro, but I know someone who bought a HDless 360 and was pissed they couldn't save games and had to buy a HD.

I believe pre-NXE arcades are the only ones that had problems due to having to have NXE installed on the HDD. Thus MS had the special program for Arcade owners at the time to pick up a specially priced 20 gig HD for 20 USD.

All Arcades MS shipped after NXE included enough flash onboard to hold the NXE as well as saves. Prior to this they all shipped with memory carts for saves.

Regards,
SB